Nobel Laureate Alvin Roth on Game Theory and Market Design

Matching markets are markets where prices cannot do all the work because participants cannot simply choose what they want even if they can afford it—they also need to be chosen by others. Examples include labor markets (where universities compete to hire professors) and college admissions (where students must be admitted by institutions). These markets function similarly to marriage, where individuals cannot unilaterally choose their spouse but must also be selected by another party.

Matching markets address allocation of indivisible goods (like organs, dorm rooms, parking spaces, school seats) when money cannot clear the market due to institutional, legal, or timing constraints. From a mechanism design perspective, matching clearing houses can be designed to make these markets function better. The lecture focuses on models without money, where matching theory becomes essential for achieving efficient allocations.

In commodity markets, prices clear the market by finding equilibrium where supply equals demand. In matching markets, prices don't determine who gets what—both parties must be chosen by each other. Stanford doesn't set tuition high enough to limit enrollment; they set it low enough that many want to attend, then select from applicants.

Matching markets differ fundamentally from traditional commodity markets where anonymous price discovery determines transactions. In matching markets, participants cannot simply purchase what they want—they must also be chosen by others. Examples include college admissions, labor markets, and marriage. Successful marketplaces must perform three essential functions: making the market thick by bringing sufficient participants together, dealing with congestion that arises from success, and ensuring safety through trust-building mechanisms. The American medical school admissions market was redesigned using a centralized computerized clearing house after unraveling caused early offers and strategic behavior. Students now submit ranked preferences alongside employers, with a computer algorithm matching everyone simultaneously. Similarly, New York City redesigned its school choice system to prevent strategic manipulation, ensuring equal likelihood of getting any choice regardless of ranking.

Market failure occurs when market transactions impose costs on third parties (social costs) that are not taken into account by the parties to the transaction. When these social costs exceed the benefits received by the transacting parties, the market transaction makes society worse off rather than increasing prosperity. This concept justifies government regulation as a corrective mechanism.

Market failures occur when one of the parties involved in an exchange engages in deceptive practices (giving 'cat for rabbit' - providing something different from what was promised). For example, a seller might deceive a buyer by selling a cat instead of a rabbit, or a buyer might use counterfeit money. However, when both parties act ethically and exchanges are voluntary, the result is an increase in well-being for all involved. Market failures are thus related to ethical behavior rather than the exchange mechanism itself.

Some markets don't develop despite apparent demand and supply (like body parts markets) due to ethical and legal considerations. Economic inequality creates situations where vulnerable populations might be exploited in markets for essential goods. Societies often establish ethical boundaries (like keeping organ donations as gifts rather than economic exchanges) to prevent exploitation of disadvantaged groups. These boundaries reflect societal values about what markets should and should not cover.

Market failure refers to situations where market transactions are not efficient or complete. Some goods and services, such as fireworks displays, cannot be provided by private markets alone and require government intervention. This concept explains why markets are not always self-sufficient.

According to Heath (2014), the justification of the market is that it produces efficient outcomes, but this only happens when the conditions of perfect competition are obtained such as perfect information, no market power, and no barriers to entry or exit. On the market failures approach, these conditions become the sources of ethical rules for market actors. However, it is difficult to select any one normative framework for applying it to a range of issues encountered in various situations in life.

The Gale-Shapley algorithm, developed by David Gale and Lloyd Shapley and awarded the 2012 Nobel Prize in Economic Sciences, provides an efficient solution to the stable matching problem where n men and n women each have preference lists over the opposite sex. The algorithm works through a deferred acceptance process where one side (typically men) proposes to their highest-ranked available partner, and the other side (women) accepts or rejects proposals based on their own preferences, ensuring termination within O(n²) iterations. The algorithm guarantees producing a stable matching—a perfect matching with no blocking pairs—and importantly, every man ends up with their best possible valid partner under stability, while every woman receives their worst possible valid partner. This elegant algorithm demonstrates how algorithmic thinking can solve complex real-world allocation problems efficiently.

A stable matching is a matching where no unstable pair exists. An unstable pair is a man and woman who are not matched to each other but prefer each other to their designated matches. The Gale-Shapley algorithm finds a stable matching: each man proposes to the first woman on his list; women collect proposals and reject all but the top proposer; rejected men go down their list and propose again; this iterates until every woman has exactly one proposer. This algorithm, developed by David Gale and Lloyd Shapley, always finds a stable matching. It has been applied to workers to jobs, colleges to students, and medical residents to hospitals.

The Gale-Shapley algorithm is a stable matching algorithm that works by having one group of people (proposers) propose to members of another group (acceptors) based on preference lists; the algorithm guarantees finding a stable matching where no two pairs would prefer to swap partners, with the proposing side always getting their best possible stable match and the accepting side getting their worst possible stable match, and the algorithm runs in O(n²) time complexity.

The stable matching problem involves finding a perfect matching between two equal-sized sets (such as doctors and hospitals) where no pair would prefer to deviate from their assigned matches; the deferred acceptance algorithm solves this by having unmatched doctors sequentially apply to their top-choice hospitals, which tentatively accept or reject based on preference rankings, with acceptances deferred until the end of the algorithm to ensure stability and guarantee termination within n² iterations.